back to home

Tropical Atlantic climate variability and change

The ITCZ is a tropical band of atmospheric deep convection band and plays a pivotal role as the driver of the tropical atmospheric circulation. The first-order physics of Atlantic ITCZ latitudinal positioning is relatively simple: it responds to the latitudinal gradients in sea surface temperature that set up the atmospheric surface pressure gradients, driving an atmospheric boundary layer circulation that positions the ITCZ. The positioning is exquisitely sensitive to changes in the sea surface temperature gradient – a change of only around 1°C between 15°N and 15°S in the tropical Atlantic shifts the ITCZ hundreds of kilometers. The tropical Atlantic ITCZ climate is therefore highly susceptible to external influences, and in fact is strongly influenced by climate variations originating from the North Atlantic and the ENSO. The densely populated northeast Brazil region – whose rainfall is essentially controlled by the Atlantic ITCZ – periodically suffers catastrophic drought as a result.

With Yochanan Kushnir and Steve Zebiak, we investigated an apparent nonstationarity in the ENSO teleconnection to March-May Northeast Brazil rainfall (Chiang et al. 2000), with weak or no association during the 1940’s 50’s and 70’s, but strong association over the last 20 years (figure 1). We proposed that this nonstationarity is tied to the nonlinearity in the convective response in the eastern equatorial Pacific to anomalous SST under it, combined with the fact that the El Ninos over the last 20 years tended to last into the March-May period whereas those in the 1940’s-70’s do not. Figure 1. 21 year sliding window correlation between MAM nino3 and Northeast Brazil rainfall in April-May (blue line), and between Jan-Feb cross-equatorial SST gradient and April-May NE Brazil rainfall (red line). The dashed line is the 5% (two sided) significance level. The bars are the number of April-May nino3 events above 28°C in a 21 year sliding window (y-values on the right).

With Yochanan Kushnir and Alessandra Giannini, we investigated the spatial-temporal nature of tropical Atlantic climate and ITCZ interannual variations. In particular, we estimated the relative influences of the ‘anomalous Walker’ mechanism of ENSO and the tropical Atlanitic cross-equatorial SST gradient on the tropical Atlantic ITCZ from observational analysis(Chiang et al. 2002); and, we investigated the same from Atmospheric GCM simulations (Giannini et al. 2001). We found that the anomalous Walker circulation primarily acts to reduces Atlantic ITCZ rainfall, whereas its positioning is controlled primarily by the SST gradient. We furthermore proposed a physical explanation for the boreal spring preference for interannual variability of the ITCZ.

The robust nature and sensitivity of Atlantic ITCZ variations suggested that its underlying mechanism is more broadly applicable to climate variability and change, a hypothesis I pursued during my tenure as a UCAR postdoctoral fellow. In collaboration with Daniel Vimont (University of Wisconsin, Madison), we showed that similar mechanism for ITCZ variability exists in the tropical Pacific, confirming its universality and challenging the notion that ENSO is the only climate variability of consequence in that region (see the “Meridional Modes” description). And, in another collaboration with David Battisti and Michela Biasutti at the University of Washington, we pushed the analogy into the paleoclimate realm by showing that the mechanism for interannual ITCZ displacement may provide a plausible explanation for long-term (century to millennial) shifts in the Atlantic ITCZ inferred from paleoproxy records (Chiang et al. 2003).

Figure 2. Last Glacial Maximum (LGM) simulations with the CCM3-slab ocean model. Top: Annual mean SST and surface wind anomalies (left), and precipitation anomalies (right), applying land ice cover, carbon dioxide, and orbital changes during LGM. Bottom: same as top, but for a simulation only using LGM land ice cover. These simulations show that the Atlantic ITCZ shifts in a manner similar to the interannual variability in the Atlantic ITCZ, and that for the LGM simulation the shift comes about primarily as a response to the land ice cover changes.

I am continuing with the pursuit of understanding marine ITCZ changes using the tropical Atlantic ITCZ variability as a model (my work on high-to-low latitude climate interactions – see the website description - is partly based on the Atlantic paleo-ITCZ study). Current and future projects include investigating marine ITCZ changes under global warming scenarios, and examining tropical Indian ITCZ interannual variability.